**** BEGIN LOGGING AT Wed Feb 27 10:59:57 2008 Feb 27 11:03:29 hi all, I created a qemux86 kernel and rootfs with OE, but I'm not clever enough to boot them in qemu. most probably I just lack a bootloader yet? Feb 27 11:04:06 you need poky-qemu script Feb 27 11:04:08 when I do $ qemu foo.bin foo.ext2, qemu says it can't read from hda Feb 27 11:04:17 I tried that too Feb 27 11:04:41 it seems to freeze qemu Feb 27 11:05:05 I get to the >qemu prompt, but i can't input anything Feb 27 11:06:13 hm ok now its not freezing qemu Feb 27 11:06:27 it says uncompressing linux...ok.... Feb 27 11:06:37 but then: BUG: Int 6: CR2 0000.... Feb 27 11:07:19 mrt2k7: qemu -kernel zImage -std-vga -hda image.ext2 -usb -usbdevice wacom-tablet --append "root=/dev/hda" Feb 27 11:07:56 did that... Feb 27 11:08:01 but i have a bzImage Feb 27 11:08:05 could that be a problem? Feb 27 11:08:09 it can be bzImage Feb 27 11:08:12 ok Feb 27 11:08:29 hmmm Feb 27 11:09:06 theres a warning at booting: "WARNING: Ancient bootloader, some funct. my be limited!" Feb 27 11:10:13 hrw: did you see the bug 3879? Feb 27 11:10:33 lumag: sorry but I lack time to watch OE progress/bugs Feb 27 11:10:56 interestingly I can boot the poky-image blinky Feb 27 11:10:58 hrw: a baby? How is it to be a father? :) Feb 27 11:11:41 lumag: baby is one thing. I also bought car recently and have important projects in work queue Feb 27 11:12:06 * lumag also has nearly no spare time lately. Feb 27 11:12:13 hrw: ok Feb 27 11:17:37 re Feb 27 12:02:11 re Feb 27 12:02:19 I can boot a vmware kernel in qemu Feb 27 12:02:47 but it cant find the rootfs...i'm appending root=/dev/hda Feb 27 12:03:24 "unable to mount root fs on unkown-block(0,0)" Feb 27 12:03:54 is 0,0 this grub terminology, like hd(0,0)? Feb 27 12:04:11 03rpurdie 07org.oe.dev * rcb2e1d76... 10/ (1 site/common-glibc): site/common-glibc: Set ac_cv_type_uid_t since autoconf detects it incorrectly Feb 27 12:04:16 03rpurdie 07org.oe.dev * r5934a535... 10/ (3 files in 3 dirs): libsdl: Remove PATH setting from sdl.m4, it breaks things Feb 27 12:04:24 03rpurdie 07org.oe.dev * rb8e6db78... 10/ (4 files in 3 dirs): prboom: Remove incorrect sdl parameter from EXTRA_OECONF, m4 patch and custom configure function since sdl.m4 now works properly Feb 27 12:04:28 i can loop-mount the ext2 image without problems Feb 27 12:04:30 03rpurdie 07org.oe.dev * reefe1e61... 10/ (3 files in 3 dirs): frodo-4.2: Remove bogus configure option and patch since sdl.m4 is fixed Feb 27 12:21:07 how does anyone measure the read/write speeds of filesystems Feb 27 12:21:13 shri: dd Feb 27 12:21:53 shri: dd if=/dev/zero of=/big/file bs=2048 count=1000000000000 Feb 27 12:21:57 Then just send it a sigusr1 Feb 27 12:23:45 something better than dd, which measure r/w speeds over a long period of time, dd may be inaccurate Feb 27 12:24:03 i am trying to measure r/w speeds of my mmc driver Feb 27 12:24:35 i tried using iozone, but i dont know how to interpret the results Feb 27 12:24:40 That's why you use a large file that's substantially bigger than your RAM, so caching the final part won't distort the readings Feb 27 12:28:15 shri: iozone, bonnie++ etc? Feb 27 12:30:01 i tried iozone having trouble interpreting the results Feb 27 12:30:46 I never used it Feb 27 12:31:12 I confess to using bonnie++ Feb 27 12:38:13 the problem is interpreting the results takes time, you get all sorts of data Feb 27 12:53:15 Could anyone tell me where STAGING_DIR_NATIVE is set please? Feb 27 12:54:29 tcooksey_: check bitbake.conf Feb 27 12:54:49 cd OE; grep STAGING_DIR_NATIVE . -R | less Feb 27 12:54:49 thx Feb 27 12:54:53 generic way :p Feb 27 12:59:14 Jin^eLD: ynezz: thx... Am using Gumstix's OE collection - which seems to be a bit out-of-date and doesn't set STAGING_DIR_NATIVE Feb 27 12:59:58 hm, I don't have any gumstix yet Feb 27 13:00:00 I'm not familiar with it, sorry Feb 27 13:02:32 Does anyone know why qmake & friends are split out into seperate packages? Feb 27 13:03:03 because there are packages using qmake, but not uic or moc Feb 27 13:03:12 likewise, there are packages using uic and moc, but not qmake Feb 27 13:03:17 etc. Feb 27 13:04:26 fair enough, bt they could be built as part of the qt* packages? Just because they've been built doesn't mean they have to be used? Feb 27 13:05:06 no, that doesn't make sense Feb 27 13:05:13 they would step on each other's toe in STAGING then Feb 27 13:05:20 qmake provides qmake Feb 27 13:05:23 uicmoc provides uic and moc Feb 27 13:05:25 qt provides qt Feb 27 13:05:43 since qt DEPENDS on uicmoc and qmake, there should be no problems Feb 27 13:05:58 what's the actual problem you're facing? Feb 27 13:06:01 qt provides qmake, uic * moc? Feb 27 13:06:19 hi all, will OM apply as a mentor organisation for the gsoc? Feb 27 13:06:23 I'm writing qt/embedded 4.4 recipies Feb 27 13:06:49 oops...wrong window Feb 27 13:07:05 good, then you can build qt/embedded which in turn DEPENDS on qmake and uicmoc IIRC Feb 27 13:07:39 well, 4.4.0-beta1 recipies to be precice. The configure scipt has changed quite a bit in 4.4 and the OE patches to disable qmake etc from being built don't apply Feb 27 13:08:05 ah. could you patch them? Feb 27 13:08:13 patch the patch? Feb 27 13:08:16 :-) Feb 27 13:08:19 yea Feb 27 13:08:32 feels... ugly. :-( Feb 27 13:08:37 well Feb 27 13:08:44 generating new patches as a result of course Feb 27 13:09:13 could do. The other problem is that qt 4.4 needs qmake from qt 4.4 to build Feb 27 13:09:16 generally, qt's buildsystem still is pretty ignorant about cross compiling. they could do us a favour if they would not assume that they generate their own tools during the build process Feb 27 13:09:41 tcooksey_: then you need to update qmake2 Feb 27 13:09:43 g'day mickeyl Feb 27 13:09:50 hi pb_ ! Feb 27 13:09:56 & mocuic Feb 27 13:10:02 tcooksey_: if necessary, yes. Feb 27 13:10:16 and do it in a way that doesn't interfere with older versions Feb 27 13:10:17 however i would be surprised if they have changed so much Feb 27 13:10:34 we compile uic, moc, qmake from qt 4.x, x being only slightly less than 4 Feb 27 13:10:41 mickeyl: my gpe-image build has completed successfully :-) Feb 27 13:10:54 sadly mythfront-image has failed due to some openchrome problem, but I can fix that one. Feb 27 13:10:57 so if they did a whole lot of incompatible changes, I'd be very surprised if they did not update the major version number Feb 27 13:11:02 pb_: oh, that's good news! congrats Feb 27 13:12:01 I'll persevere, but I still don't understand why qt can't provide qmake, uic & moc Feb 27 13:12:09 mickeyl: heh, ueberhacker zecke did most of the hard work Feb 27 13:12:26 tcooksey_: because Feb 27 13:12:29 a) we have multiple qt versions Feb 27 13:12:35 b) people are using qmake but not qt Feb 27 13:13:19 c) we have to support multiple qmake versions Feb 27 13:14:00 we need to ensure we are using the same tools Feb 27 13:14:18 we can not allow qt using different tools as applications basing on qt Feb 27 13:14:20 this will lead to chaos Feb 27 13:14:24 so our way is much cleaner Feb 27 13:14:56 trust me, we came from having qt provide everything... it didn't work. splitting it up was painful, but it works very solid nowadays Feb 27 13:15:33 note also that we need to support build machines without X installed Feb 27 13:15:48 which is why we actually build qmake from qt/e Feb 27 13:15:51 qmake-native, that is Feb 27 13:16:34 ok.. will attempt to continue with current structure. I've spoken to trolls about splitting qmake & friends into seperate packages and their point was that qt needs the version of the tools it was shipped with to build itself, so why not just bundle them all together Feb 27 13:16:50 yes, that's the general attitude of trolltech Feb 27 13:16:58 being very ignorant about cross compiling Feb 27 13:17:09 i'm sorry to say that but they have no clue about build systems Feb 27 13:17:24 the ./configure alone is a mess Feb 27 13:17:43 well, cross-compiling qtopia core seems to work ok. They have political issues for cross-compiling qt/x11 Feb 27 13:17:56 brisbane does not improve the situation by forking their own product (qmake) Feb 27 13:18:30 true, but OE does not care about political issues. we care about reproducing builds Feb 27 13:18:52 please try to base on what we have now Feb 27 13:19:01 and we'll help you once you run into problems Feb 27 13:19:19 * XorA hides under the bed at all this qt/qtopia talk Feb 27 13:19:24 the current resident qt expert is on his way to Taipei but will probably be online again soon Feb 27 13:19:37 z? Feb 27 13:19:39 ya Feb 27 13:19:57 i did all the initial work in splitting that stuff up Feb 27 13:20:07 but lost interest after a couple of years with nothing but problems with TT Brisbane Feb 27 13:20:11 he seems pretty busy the few times I've spoken with him. Feb 27 13:20:12 *shrug* Feb 27 13:20:31 true Feb 27 13:20:51 but he's going to be paid to resolve qtopia build problems now Feb 27 13:20:57 so this is absolutely on topic ;) Feb 27 13:21:48 LOL... while porting qtopia to qt/x11 will solve these kind of build issues... not sure it really counts! Feb 27 13:22:03 *nod* Feb 27 13:22:10 the whole future of qtopia is completely unclear if you ask me Feb 27 13:22:15 but we'll wait and see Feb 27 13:22:36 no comment. :-) Feb 27 13:22:42 of course ;) Feb 27 13:23:32 i just know that getting qtopia into OE would very likely refreshen developer interest in it Feb 27 13:23:41 so it's very important for people who care about it Feb 27 13:23:57 and that's why I'm encouraging everything interested Feb 27 13:24:02 everyone, even Feb 27 13:24:22 [having my UI agnostic OE hat on] Feb 27 13:24:29 btw: there's some patches I'm not sure are doing the right thing in qt4... *-host-includes seems odd Feb 27 13:25:22 hmm Feb 27 13:25:24 Do you plan to use Qtopia on Gumstix? Feb 27 13:25:27 looks perfectly good to me Feb 27 13:25:38 why would you want to include the hosts's freetype headers over the ones in OE? Feb 27 13:25:39 - TRY_INCLUDEPATHS = /include /usr/include $$QMAKE_INCDIR $$QMAKE_INCDIR_X11 $$INCLUDEPATH Feb 27 13:25:39 + TRY_INCLUDEPATHS = $$QMAKE_INCDIR $$QMAKE_INCDIR_X11 $$INCLUDEPATH Feb 27 13:25:41 mickeyl: Error: "TRY_INCLUDEPATHS" is not a valid command. Feb 27 13:25:48 makes much sense to me Feb 27 13:25:53 we have to forbid using host includes Feb 27 13:26:35 ah yes... got my -/+ mixed up when looking at the diff... Feb 27 13:27:23 EUGHHH.... that's disgusting... configure _forces_ non-prefixed include paths... Feb 27 13:28:39 ok, not discusting... it's only used when not cross-compiling Feb 27 13:28:48 huh Feb 27 13:29:15 actually i don't know whether we are telling configure that this is a cross-compile or not, because IIRC we had rpoblems with configure's idea of a cross compile. Feb 27 13:29:34 but i have not looked at that for over a year Feb 27 13:31:43 most of the OE patches seem to be there to 1) split out tools and 2) allow cross-compiling of qt/x11 Feb 27 13:33:03 mickeyl: I think we tell it its compiling on TARGET to get around qtopia trying to be too smart, I had the problem that this did not work on qtopia-phone Feb 27 13:33:22 mickeyl: so qtopia-phone knows it is cross compiled, qtopia-core doesnt Feb 27 13:33:39 righto Feb 27 13:34:13 mickeyl: either that or I misunderstood the whole process (perfectly plausible) Feb 27 13:35:22 is there anyway of getting my 4.4 recipie to replace 0003-no-tools.patch Feb 27 13:35:25 mickeyl: BTW your rotation problem on gta02 should be fixed by checkin I did to OE this morning Feb 27 13:35:54 tcooksey_: sure. just make a directory that has your version number encoded Feb 27 13:36:08 this will be preferred over a directory with only ${PN} Feb 27 13:36:15 e.g. name it ${PN}-${PV} Feb 27 13:36:28 XorA: right, the git version ,thans Feb 27 13:36:35 XorA: i'll rebuild from scratch later today Feb 27 13:36:44 XorA: BT seems to work here as well, but i have problems with WiFi Feb 27 13:36:44 mickeyl: and I turned off tick tick tick in gsmd init script Feb 27 13:36:54 good Feb 27 13:36:57 mickeyl: I have no BT radio, wifi doesnt accept keys Feb 27 13:37:10 so qt4/qt-embedded-linux/0003-no-tools.patch will replace qt4/files/0003-no-tools.patch? Feb 27 13:37:30 mickeyl: and I fixed pulse s/adduser/addgoup/ :-D Feb 27 13:37:53 tcooksey_: kind of Feb 27 13:37:57 say your new recipe is called Feb 27 13:38:02 qt-embedded-linux_4.4.0 Feb 27 13:38:07 then create a directory called Feb 27 13:38:10 qt-embedded-linux-4.4.0 Feb 27 13:38:13 this will be preferred over Feb 27 13:38:15 qt-embedded-linux Feb 27 13:38:16 and Feb 27 13:38:17 files Feb 27 13:38:33 urghsd Feb 27 13:38:42 paxutils native wants to install into /usr/bin etc. Feb 27 13:38:45 *sigh* Feb 27 13:39:13 * mickeyl fixes Feb 27 13:39:58 regarding qt. I will be messing today around getting qtjambi into OE :) Feb 27 13:40:07 cool Feb 27 13:40:25 so files in qt-embedded-linux won't be prefered over files in files? I have to append the version? Feb 27 13:40:59 you don't have to Feb 27 13:41:04 it will work without Feb 27 13:41:08 ${PN} is more specific than files Feb 27 13:41:15 tcooksey_: the more specific the directory name to higher the priority, so lowest is files, then qt-embedded-linux, then qt-embedded-linux-4.4.4 Feb 27 13:41:15 i just think it's clearer Feb 27 13:41:57 righto... thx Feb 27 13:56:09 mickeyl: Next question... why patch the existing mkspecs/common/* rather than just create a new mkspec for the OE toolchain? Feb 27 13:56:54 (this is 0005-fix-mkspecs.patch I'm talking about) Feb 27 13:57:45 good question Feb 27 13:57:52 we do create own mkspecs Feb 27 13:57:55 (see qmake2-native) Feb 27 13:58:06 yeah... which makes sense Feb 27 13:58:30 this patch might either be bitrotten or due to the fact that for some reason someone could not get our mkspecs to be used Feb 27 13:59:23 one gotcha I found for mkspecs was they have to only contain a maximum of 3 - signs :-) Feb 27 14:00:18 as in "foo---bar" is an invalid token? Feb 27 14:00:51 oe-linux-g++ is ok, arm-gnueabi-linux-g++ is not Feb 27 14:01:06 * XorA meant 2 - symbols Feb 27 14:02:20 looking at qmake_base.bbclass, the makespec is passed to qmake via a command line arg, I can't see how this can not work? Feb 27 14:02:49 XorA: what do you mean when you say arm-gnueabi-linux-g++ isn't ok? Feb 27 14:03:48 tcooksey_: make a directoy called that and things go wrong :-( Feb 27 14:04:07 why would OE use such a makespec... surely oe-linux-g++ is always used as the makespec passed into qmake? Feb 27 14:06:19 ah, ok... so it will be *linux-oe-g++|*linux-uclibc-oe-g++|*linux-gnueabi-oe-g++|*linux-uclibcgnueabi-oe-g++ Feb 27 14:09:35 mickeyl: were you able to make the required changes so that busybox and tinylogin would ipkg install on the nslu2 slugos? I checked yesterday and it still won't install due to that conflict that we discussed Feb 27 14:10:10 03mickeyl 07org.oe.dev * re2f4456d... 10/ (1 packages/pax-utils/pax-utils-native_0.1.13.bb): pax-utils-native: yank do_install Feb 27 14:10:30 aadamson: unfortunately i had zero time to do it while i was on FOSDEM. sorry, this is on my agenda for today Feb 27 14:10:35 03koen 07org.oe.angstrom-2007.12-stable * r3165c781... 10/ (45 files in 12 dirs): Feb 27 14:10:35 backports from .dev: applied gcc and metatoolchain updates Feb 27 14:10:35 * http://lists.linuxtogo.org/pipermail/angstrom-distro-devel/2008-February/001901.html Feb 27 14:10:35 * http://lists.linuxtogo.org/pipermail/angstrom-distro-devel/2008-February/001902.html Feb 27 14:10:57 ah, yes, FOSDEM, no problem just saw you on here and thot I'd check... Thanks again for all the help! Feb 27 14:11:14 mhm.. still trying to build uclinux stuff... I am using global CFLAGS settings like: CFLAGS_uclinux-uclibcgnueabi_append=" -Dfork=vfork -Wl,-elf2flt ", however I figure that gcc should not be compiled using those settings Feb 27 14:11:30 is there a way to somehow say "do not apply these appends to certain packages" Feb 27 14:11:39 or do I have to add some magic to the gcc .bb recipes? Feb 27 14:25:55 hmm, oe website down? Feb 27 14:26:19 I can ping the server but no website when trying from firefox.. Feb 27 14:26:48 ah, now it came back.. must be connection issues or whatever Feb 27 14:31:01 mickeyl: XorA: The OE installs mkspecs for each libc/abi it supports and those mkspecs just inherit the common mkspecs, hence the need to patch the common mkspecs. :-) Feb 27 14:31:21 aah, right. good catch Feb 27 14:32:00 well.. technically OE cought it when it failed to build. :-D Feb 27 14:33:40 still don't understand why there needs to be a seperate mkspec for each lib/abi rather than just an uber oe-g++ mkspec, but past caring at this point. Feb 27 14:55:04 hi? can anybody help me with with my shiny new OE-rootfs and kernel? how do you make the results of bitbake task-bootstrap actually bootable? Feb 27 14:56:27 mrt2k7: what distro were you building for? Feb 27 14:57:11 usually you build a console-image or minimal-image or whatever and depending on your hardware you can use the result to boot via NFS if you're developing or flash it or whatever Feb 27 15:00:06 i build an image for vmware Feb 27 15:00:13 distro generic i think Feb 27 15:00:50 i would like to add a bootloader like lilo or grub Feb 27 15:00:55 I did not play around with it, but check your deploy directory and look what kind of image has been produced Feb 27 15:01:09 yeah theres a bzImage and a rootfs Feb 27 15:01:35 ~lart people sending e-mails with borked clocks Feb 27 15:01:35 * ibot plops people sending e-mails with borked clocks into a giant vat of herring Feb 27 15:02:13 you'd probably have to boot from some CD so you can install grub or something, I did not try running x86 OE images in vmware Feb 27 15:02:56 03mickeyl 07org.oe.dev * rfaf71c56... 10/ (10 files in 4 dirs): python 2.5.1 upgrade to 2.5.2 Feb 27 15:03:03 03mickeyl 07org.oe.dev * r442fae6a... 10/ (5 files in 2 dirs): add pyneod, remove python-fso-pyproto Feb 27 15:03:11 03mickeyl 07org.oe.dev * r32e18537... 10/ (3 files in 3 dirs): add pyneog Feb 27 15:06:11 Do we really need 17 different versions of pvrusb-mci? Feb 27 15:12:16 most likely no Feb 27 15:15:22 rwhitby: It seems slugos is the main user of pvrusb-mci? Feb 27 15:16:16 RP: that would be peripherally true, yes. i.e. it's a package that someone once wanted added to SlugOS, cause you can use a PVRUSB thingo attached to a slug. Feb 27 15:16:31 so the latest would be fine for us. Feb 27 15:16:40 and I have no way of testing that :-) Feb 27 15:17:12 rwhitby: Can you confirm that you're using 20060903 ? Feb 27 15:20:09 RP: actually, I can confirm that pvrusb-mci is listed in BROKEN_PACKAGES at the moment in slugos, and has been for some time, so we are not actually supporting it at the moment. Feb 27 15:20:22 morning Feb 27 15:20:24 so any version that builds with recent kernels would be a step forward for us. Feb 27 15:33:27 bye Feb 27 15:38:43 Crofton: Morning. Feeling any better? Feb 27 15:38:51 rwhitby: ah, right. I really don't want to get too involved in fixing lots of things atm :/ Feb 27 15:38:51 yeah Feb 27 15:38:53 slowly Feb 27 15:38:58 brain is not as foggy Feb 27 15:41:54 rwhitby: 20080210 builds... Feb 27 15:42:23 RP: feel free to remove the others then. Feb 27 15:46:42 Anyone around that feels empowered to review udev or samba patches? Feb 27 15:48:41 rwhitby: 2008 added and everything but the last 2006 dropped Feb 27 15:48:58 Crofton: Where would you like me to send you my mailing address? Feb 27 15:49:06 RP: sweet, thx. will test a build tonight. Feb 27 15:49:49 Its so easy to get sidetracked when fixing these STAGING_DIR references :/ Feb 27 15:51:21 just msg'd my emai Feb 27 15:53:40 bye Feb 27 15:54:42 Has anyone noticed the parsing speedups? Feb 27 15:55:44 The boost recipe has "PACKAGES =\n" in it :/ Feb 27 16:03:54 RP: Do I need to use svn head for that Feb 27 16:04:02 hvontres|work: yes Feb 27 16:04:04 03mickeyl 07org.oe.dev * r3dc26cf0... 10/ (3 files in 3 dirs): first shot at moko-underground-image Feb 27 16:04:10 03rpurdie 07org.oe.dev * r7496170a... 10/ (17 files in 2 dirs): pvrusb2-mci: Fix STAGING references in TARGET_LDFLAGS, add 20080210 version which builds against modern kernels Feb 27 16:04:19 03rpurdie 07org.oe.dev * r0727347c... 10/ (1 packages/pvrusb2-mci/pvrusb2-mci2.inc): pvrusb2-mci: Add missing .inc from last commit Feb 27 16:04:26 03rpurdie 07org.oe.dev * r42b4313e... 10/ (43 files in 15 dirs): pvrusb2-mci: Remove a ton of old versions Feb 27 16:06:15 RP: testing now Feb 27 16:19:52 RP: old 181 s, new 136 s for full parse of .dev :) Feb 27 16:20:14 * hvontres|work wonders what to do whith all those extra minutes :) Feb 27 16:30:48 * cbrake starts digging into the Xfbdev "no screens found" error message ... Feb 27 16:37:17 hiya... me again. :-) Having troubles with missing includes for dbus Feb 27 16:37:51 specifically, dbus-arch-deps.h does not make it into ${STAGING_INCDIR}/dbus-1.0 Feb 27 16:42:26 anyone knows how to include/use an existing .bbclass file? Feb 27 16:44:00 there's bootimg.bbclass, but docs don't say where to include it in my conf Feb 27 16:53:00 the "intltoolize" binary is provided by which OE package? Feb 27 16:53:13 intltool-native Feb 27 16:54:09 mickeyl: thanks Feb 27 16:55:13 hi all, somebody can help with klibc? Feb 27 16:56:08 I'm poking with initramfs-images and I fear the latest thesing's commit (eabi patch) broke it... Feb 27 16:57:28 ~seen thesing Feb 27 16:57:34 thesing was last seen on IRC in channel #oe, 1d 3h 26s ago, saying: 'morning'. Feb 27 17:04:15 Anyone around who feels empowered to review samba or udev changes? Feb 27 17:05:01 best put it forwards to the list Feb 27 17:05:06 Tartarus: I'll try your patches as soon as possible (over VPN) Feb 27 17:05:15 ant|work, thanks Feb 27 17:05:26 and/or attach to bugzilla Feb 27 17:06:00 mickey|sports, is attached :) Feb 27 17:06:40 Tartarus: [ebuild R ] net-fs/samba-3.0.28 Feb 27 17:06:54 Tartarus: should I expect issues from that version? Feb 27 17:07:31 ant|work, er? I didn't see 3.0.28 in OE when I did my patches Feb 27 17:07:47 noo, on the remote host (my office) Feb 27 17:07:56 ok Feb 27 17:08:10 I still don't follow, sorry :) We should ad a recipe for 3 Feb 27 17:08:21 Tartarus: I'll test smbclient Feb 27 17:08:27 .0.28 as well, and it would have the smbfs/cifs bugs if based on the other oens Feb 27 17:08:40 ok Feb 27 17:08:44 against my remote samba server on Gentoo Feb 27 17:08:49 Ah Feb 27 17:08:55 over VPN router-router Feb 27 17:08:57 the problem is how we package it up for the targets Feb 27 17:09:03 keep in mind :) Feb 27 17:09:05 but yes Feb 27 17:09:07 k Feb 27 17:09:26 you don't even need a samba server to see the problem actually Feb 27 17:09:36 mount -t smb / mount -t smbfs gets upset before even trying ot mount Feb 27 17:09:43 I see Feb 27 17:09:52 (or rather, its failing to find mount.smb/mount.smbfs and stopping there) Feb 27 17:44:30 hvontres|work: A 25% speedup is rather nice, thanks for the figures :) Feb 27 17:50:27 03rpurdie 07org.oe.dev * r2d630872... 10/ (1 contrib/powered-by-oe.svg): contrib: Add powered by OE logo Feb 27 18:23:02 03rpurdie 07org.oe.dev * r059498b6... 10/ (3 files in 2 dirs): boost: Fix packages so they build Feb 27 18:23:06 03thebohemian 07org.oe.dev * r23d83842... 10/ (1 packages/docbook-utils/docbook-utils-native_0.6.14.bb): docbook-utils-native 0.6.14: Provide yet another name for all tools (via symlink). Feb 27 18:23:11 03thebohemian 07org.oe.dev * r37c00188... 10/ (5 files in 3 dirs): xvidcap: Added new recipes. Feb 27 20:46:21 hi all Feb 27 20:46:35 hi gremlin[it] Feb 27 20:47:41 I have a question, suppose some package are available as precompiled only (.ipkg) ... what's the best way to make this package be included in an image i'm building ? Feb 27 20:47:43 hi gremlin[it] Feb 27 20:48:07 1) have a "fake" recipe that just download the prebuild.ipkg ? Feb 27 20:48:22 2) is already available some way in OE ? Feb 27 20:48:29 hi pb__ !!! Feb 27 20:49:28 gremlin[it]: I guess you could copy the .ipk files into the deploy directory by hand and then set ASSUME_PROVIDED to stop oe trying to rebuild them. Feb 27 20:50:44 mhhh if suck prebuild package isn't available in normal way in OE ... is simple (not difficult) to write a "fake" recipe i suppose ... Feb 27 20:50:54 fi suck -> if such ... Feb 27 20:51:08 yeah, that would work too Feb 27 20:51:19 I don't know of any dedicated mechanism for using prebuilt binaries Feb 27 20:52:07 should exist something but for binary archive ... not a prebuild package i thing Feb 27 21:35:41 I'm new to oe. I only used oe once to build a rootfs for gumstix. Is building for another platform just a matter of running "bitbake htcblueangel" or do I need to get some other configuration files? Feb 27 21:36:23 tre9: You probably want to change MACHINE Feb 27 21:36:51 ah, ok thanks. but nothing else to install? Feb 27 21:37:23 tre9: I'm don't know the gumstix setup well but probably not Feb 27 21:37:59 tre9: most of the top level build configs are done through the local.conf that you set up Feb 27 21:38:52 tre9: then when you do a bitbake XX it will pull the appropriate files for your machine Feb 27 21:39:01 hopefully :) Feb 27 21:41:05 my local.conf doesn't have much in it http://pastebin.ca/921201 Feb 27 21:41:32 there's some gumstix-specific settings in site.conf Feb 27 21:42:26 http://pastebin.ca/921205 Feb 27 21:42:40 did you work off the local.conf.example file in the org.oe tree? Feb 27 21:43:20 err org.openembedded.dev/conf/local.conf.sample Feb 27 21:43:35 you will see options for MACHINE Feb 27 21:43:45 like RP was referring too Feb 27 21:43:51 not sure, it seems that gumstix has done some of the configuration. Feb 27 21:44:43 which gumstiks platform are you using? Feb 27 21:44:46 MACHINE is set to "gumstix-custom-verdex" in auto.conf Feb 27 21:45:13 tre9: auto.conf is probably generated by one of the gumstix scripts. You need to change that variable Feb 27 21:45:23 yeah I'm not familiar with setting up gumstix through their scripts, I did everything by hand Feb 27 21:45:37 if you look at org.openembedded.org/conf/machine Feb 27 21:45:54 * cbrake learns once again that Xfbdev really does not like 8bpp displays ... Feb 27 21:45:59 you will find gumstix-connex, gumstix-custom-connex, and others Feb 27 21:46:13 hello cbrake Feb 27 21:48:16 RP: ok, thanks. Feb 27 21:49:44 Fique: hello Feb 27 21:50:14 cbrake: just wanted to thank you again for all the cm-x270 help Feb 27 21:50:22 Fique: sure Feb 27 21:50:31 Fique: pushing 2.6.24 support shortly :-) Feb 27 21:50:35 tre9, for the gumstix oe system you should ask in #gumstix Feb 27 21:50:47 cbrake: awesome... id be willing to help again with PCI testing, i've had it working fine for 2.6.23 Feb 27 21:56:39 ok, looks like I set MACHINE = "htcblueangel" and it will get other conf from org.openembedded.snapshot/conf/machine/htcblueangel.conf Feb 27 21:57:34 Fique: good point. Refresh my memory -- was it just a few defconfig changes, or was more required to get PCI going? Feb 27 21:57:46 Crofton: thanks for the tip. I'll check there too. Feb 27 21:58:12 they use a snapshot of the OE metadata and have local fixes for gumstix cards Feb 27 21:58:56 will that be a problem for me when I build for a non-gumstix machine? Feb 27 21:59:09 yes Feb 27 21:59:21 because we can't help you with problems :) Feb 27 21:59:27 What are you trying to do? Feb 27 21:59:57 I want to build for a handheld device (HTC blueangel) Feb 27 22:00:11 why did you start with the gumstix stuff then? Feb 27 22:01:55 I also have a gumstix platform I need to build. Feb 27 22:01:59 ah Feb 27 22:02:06 for same app ? Feb 27 22:02:46 * Crofton hopes sakoman reads this conversation Feb 27 22:02:56 I need the whole rootfs. no custom apps. Feb 27 22:03:18 I need to set QTDIR for a library that needs the qt4 headers. Can I safely set this to some dir within ${STAGING_INCDIR}? Feb 27 22:03:27 but you run on gumstix and htc machine Feb 27 22:03:50 if the files in the gumstix snapshot are OK, you should be fine Feb 27 22:03:55 yes. Feb 27 22:04:08 basically try it and see what happens Feb 27 22:04:30 we get along well with the gumstix guys, so we can do what we can Feb 27 22:05:23 rschuster: There are qt classes which set that variable iirc Feb 27 22:05:44 rschuster: It seems to default to qt2 in bitbake.conf though, I was just looking at that Feb 27 22:07:09 RP: hm, some bbclasses reference QTDIR .. Feb 27 22:07:21 chinook-compat produces libstdc++6_3.4.4+csl-arm-2005q3, ITOS2008 has libstc++6_3.4.4cs2005q3.2-5.osso8. is there a way to build kobodeluxe so that I don't have to --force-depends it on the n810? Feb 27 22:08:05 Crofton: ok. sakoman's tutorial was easy to follow. gave me courage to try oe. :) now I need to take the next step. Feb 27 22:08:06 pH5: unfortunately not yet. there was a proposal on the list to make version numbers overridable. that would fix it Feb 27 22:09:08 well welcome aboard Feb 27 22:09:58 rschuster: I think that proposal won't work due to the weird locations the version ends up :/ Feb 27 22:10:01 He is trying to get up to the Angtstrom stable branch, which will help some by giving us a route from .dev to stable to him Feb 27 22:10:24 rschuster: Its an interesting idea though Feb 27 22:11:03 rschuster: I was the one who fixed a large number of package renaming bugs, the thought of doing it again for version numbers scares me ;-) Feb 27 22:20:12 I see, that's a pity. What's the alternative? Setting PV_chinook-compat="3.4.4cs2005q3.2-5.osso8" in gcc_csl-arm-2005q3.bb? Feb 27 22:21:14 #oe is very helpful. Thanks. I'll probably be back with more questions. Feb 27 22:21:19 Doesn't sound too clean, we probably don't want to litter core recipes with distro overrides :) Feb 27 22:25:11 cbrake: sorry for the delay, went out to pick up lunch. But its defconfig changes, one file addition thats missing in the tarball, and finally one function parameter change in an IDE driver file Feb 27 22:25:29 cbrake: and thats it Feb 27 22:28:05 * * OE Bug 3889 has been created by  Feb 27 22:28:07 * * ppp-dialin-0.1-r4-do_install Feb 27 22:28:08 * * http://bugs.openembedded.net/show_bug.cgi?id=3889 Feb 27 22:29:23 pH5: no not a good idea. I still think we need the version override thing :| Feb 27 22:30:51 Fique: ahh, yes just found your email you sent some time ago Feb 27 22:31:21 Fique: I'll try to include these changes in the 2.6.24 stuff I push tomorrow. If you have any updates or anything you want changed, feel free to send it. Feb 27 22:31:41 cbrake: please test though, I can't verify anything with 2.6.24 Feb 27 22:32:00 cbrake: or i'll test it myself when its pushed Feb 27 22:32:05 * * OE Bug 3890 has been created by  Feb 27 22:32:07 * * libxinerama-1.0.2-r3-do_populate_staging Feb 27 22:32:09 * * http://bugs.openembedded.net/show_bug.cgi?id=3890 Feb 27 22:32:16 Fique: will do -- at least that it does not break any of more normal usage patterns Feb 27 23:08:56 03tmbinc 07org.oe.dreambox * r56c92928... 10/ (3 files in 2 dirs): dreambox-dvb-tools{,-v3}: fix qmake file to not link qte Feb 27 23:09:01 03thebohemian 07org.oe.dev * rb2b754e5... 10/ (6 files in 2 dirs): Feb 27 23:09:01 classpath-minimal 0.97: Added recipe. Feb 27 23:09:01 classpath 0.97: Added recipe. Feb 27 23:09:01 classpath: Added proper RPROVIDES for ${PN}-common. Feb 27 23:10:08 RP: Error: packages/pvrusb2-mci/pvrusb2-mci-20080210/Makefile.patch not found. Feb 27 23:10:30 rwhitby: oops Feb 27 23:10:54 don't you just hate mtn add ;-) Feb 27 23:11:16 rwhitby: yes ;-) Feb 27 23:15:39 thesing: I had some minor issues with klibc yesterday (the eabi patch was not applied) Feb 27 23:16:20 thesing: then there were bigger problems with busybox-static Feb 27 23:17:10 hey philippe are you there? Feb 27 23:23:15 RP: bitbake seems stalling on "NOTE: Executing runqueue". /tmp is empty. Feb 27 23:23:29 RP: fresh build Feb 27 23:24:24 RP: http://www.pastebin.ca/921338 Feb 27 23:25:36 RP: 10818 andrea 20 0 182m 128m 2216 R 100 6.4 6:05.78 python Feb 27 23:30:19 ant_: no idea :/ Feb 27 23:30:43 RP: python is eating 1367M virt Feb 27 23:31:05 ant_: Its obviously spinning out of control in some circular loop Feb 27 23:31:21 RP: was working until I wiped /tmp Feb 27 23:31:29 ant_: What were your building? Feb 27 23:31:34 s/your/you/ Feb 27 23:31:39 initramfs under uclibc Feb 27 23:32:03 then I prefer restart from scratch Feb 27 23:36:18 RP: if I kill it with ctrl+C python should spit out some traces... Feb 27 23:36:42 ant_: It does it here too which should help the debugging Feb 27 23:38:30 Its spinning in check_stamps() :/ Feb 27 23:45:10 03rpurdie 07org.oe.dev * r4e74a697... 10/ (1 packages/pvrusb2-mci/pvrusb2-mci-20080210/Makefile.patch): pvrusb2-mci: Add missing patch Feb 27 23:50:54 03rpurdie * r1022 10/ (4 files in 4 dirs): fetch/__init__.py: Make fetcher timestamp updating non-fatal when permissions don't allow updates Feb 27 23:51:33 03rpurdie * r1023 10/ (2 files in 2 dirs): runqueue.py: Fix recursive looping issue Feb 27 23:52:41 ant_: That should fix it Feb 27 23:55:01 indeed, thx Feb 27 23:56:30 RP: about uclibc build, I think it somehow contaminated my staging Feb 27 23:56:53 ant_: entirely possible Feb 27 23:57:18 RP: ANGSTROM_MODE=uclibc IMAGE_FSTYPES="cpio.gz" bitbake -crebuild initramfs-bootmenu-image built Feb 27 23:57:48 RP: with errors :( Feb 27 23:58:03 RP: initramfs-kexec-images built fine :) Feb 28 00:02:11 let see tomorrow, gn Feb 28 00:04:18 I wrote a package for linux-uvc (a driver for UVC webcams).. Could someone with dev access commit it for me? Feb 28 00:24:27 03rpurdie 07org.oe.dev * r630bb2aa... 10/ (20 files in 2 dirs): Feb 28 00:24:27 gcc: Enable sysroot options, hacking around path layout issues for now. This Feb 28 00:24:27 fixes the sdk version to use the correct sysroot which is important for ease of Feb 28 00:24:27 use of SDKs and standalone toolchains. This change means a gcc rebuild Feb 28 00:24:27 unfortunately Feb 28 01:07:49 Anyone know what happened to gtk+-directfb ? Feb 28 01:10:12 ERROR: '['/home/slug/angstrom/openembedded/packages/gtk+/gtk+-directfb_2.10.14.bb']' RDEPENDS/RRECOMMENDS or otherwise requires the runtime entity 'gtk+-directfb' but it wasn't found in any PACKAGE or RPROVIDES variables of any buildable targets. Feb 28 01:41:31 03rwhitby 07org.oe.dev * rccc4dd46... 10/ (4 files in 4 dirs): linux-ixp4xx: Enable V4L1_COMPAT for pvrusb2-mci Feb 28 02:45:04 * * OE Bug 3891 has been created by  Feb 28 02:45:06 * * obexpush-1.0.0-r3-do_compile Feb 28 02:45:08 * * http://bugs.openembedded.net/show_bug.cgi?id=3891 Feb 28 02:51:20 sakoman: i posted a question in the icecast forum. Karl, one of the ices developers, had an interesting response Feb 28 02:51:22 http://icecast.imux.net/viewtopic.php?p=12067#12067 Feb 28 02:51:52 oops Feb 28 02:51:55 http://icecast.imux.net/viewtopic.php?p=12068#12068 Feb 28 06:10:54 03xjqian 07org.oe.dev * r2467bb13... 10/ (3 files in 3 dirs): netsurf_svn: fix srcrevs Feb 28 06:11:01 03xjqian 07org.oe.dev * r3802d5aa... 10/ (3 files in 3 dirs): ffmpeg_svn: fix SRCREV Feb 28 08:05:33 g'morning all! Feb 28 08:49:09 re Feb 28 08:55:57 rwhitby: Its missing pango-directfb. Some metadata changes made the error more obvious but the problem was always there Feb 28 08:56:39 morning Feb 28 09:00:24 hi hrw! Feb 28 09:00:42 how's your daughter ? Feb 28 09:00:55 fine Feb 28 09:01:02 tomorrow should be at home Feb 28 09:01:45 nice ! Feb 28 09:10:52 03rpurdie * r1024 10/ (4 files in 4 dirs): runqueue.py: Add BB_SCHEDULER and BB_STAMP_POLICY variables Feb 28 09:24:27 what do I write in my .bb file other than file://makefile \ to specify parameters which I would normally do like make -s TARGET_ARCH=.... ? Feb 28 09:27:58 pbne04: you would probably write your own do_compile() Feb 28 09:30:57 where can I find a guide or tutorial for this? Feb 28 09:31:45 pbne04: the easist would be to have a look at some .bb files in the packages directory Feb 28 09:32:46 grep for do_compile, you'll find some .bb recipes to see how its being done, the bitbake manual might also be helpful Feb 28 09:33:07 Jin^eLD: thanks Feb 28 09:33:17 np Feb 28 09:39:43 RP: ping :> Feb 28 09:51:25 hey guys, i have a working oe-build now, thx to the help of many of you ;) Feb 28 09:52:19 anyways i was just asking myself if someone has build generic-uclibc successfully? Feb 28 09:53:36 Jin^eLD: pong Feb 28 09:53:39 you mean generic distro? I base mine on Angstrom and I did some uclibc builds Feb 28 09:54:18 RP: how can I omit the global appends for certain packages? Feb 28 09:54:32 allthough after numerous retries I am not yet sure if that is actually the problem :) Feb 28 09:55:05 I get the impression that uclinux-uclibc that Koen was using for bfin did not work for gcc, i.e. - it did not know that string Feb 28 09:55:37 Jin^eLD: You can't really disable something like that on a per package basis Feb 28 09:55:47 RP: siteinfo.bbclass resolves those strings and decides if I go with uclibc or glibc and so on? Feb 28 09:56:03 yeah, the generic-distro - is your uclibc build substantially smaller? Feb 28 09:56:10 Jin^eLD: You can override the string entirely on a per package basis Feb 28 09:56:18 (cmp to glibc) Feb 28 09:56:42 tomimo: I did not compare it, I did it only to compile some specific packages that I used as an addon to some other existing rootfs Feb 28 09:56:46 Jin^eLD: siteinfo.bbclass decides which site files to use to give hints to autofoo Feb 28 09:56:50 crap, IRC nick completion, tom_ I mean Feb 28 09:57:18 RP: somehow after changing my TARGET_OS to uclinuxeab it keeps selecting glibc instead of uclibc Feb 28 09:57:28 and uclinuxeabi seems to be the "correct" token for gcc Feb 28 09:57:57 Jin^eLD: What is selecting glibc? Feb 28 09:58:11 not sure yet, trying to figure that out Feb 28 09:59:03 too bad Koen is not around, I do not know how far he got with his bfin attempts, I only see some things here and there and I am wondering if he got further than I am now Feb 28 09:59:22 Jin^eLD: mail him or jabber Feb 28 09:59:48 hrw: where can I find his jabber addy? Feb 28 10:00:06 what is angstrom in a short way? is it just for handhelds? Feb 28 10:00:33 because i'm building for qemu/vmware, so i assume its not so important for me? Feb 28 10:00:35 tom_: angstrom is a maintained distro, at least imho it's a good starting point if you want to customize your own Feb 28 10:00:58 k thx Feb 28 10:01:02 will have a deeper look Feb 28 10:01:16 i startet with generic, and it's really damn small Feb 28 10:01:19 and runnable :) Feb 28 10:01:30 6,7mb including qemu for windows :)) Feb 28 10:01:37 :) Feb 28 10:01:40 compressed Feb 28 10:02:48 can i think of OE like debian as a whole and angstrom something like etch? Feb 28 10:04:04 tom_: not quite, I'd say OE is something that allows you to build a distro for various architectures, so angstrom in that sense would be like debian Feb 28 10:05:16 and if you want the etch/sid comparison - there is a stable and unstable branch of OE and also of angstrom Feb 28 10:05:41 ok...and when i do DISTRO="angstrom..." i can build it for various MACHINEs? Feb 28 10:06:12 and if continue the comparisons then you could build other distros too Feb 28 10:06:22 yes Feb 28 10:06:35 ok that sounds good...thx Feb 28 10:06:50 last question: anybody working with icecc? Feb 28 10:07:04 because building takes quite long even on weird hardware Feb 28 10:07:51 <_diego_> http://www.openembedded.org/wiki/IceCC Feb 28 10:08:39 thx, read that yet...i was wondering if its some kind of stable yet Feb 28 10:09:04 <_diego_> i know of people using it Feb 28 10:10:20 k thx Feb 28 10:11:14 tomimo: if you have several cores and fast drivers, and use parallel build options it's going quit efast Feb 28 10:12:35 s/drivers/drives Feb 28 10:12:47 yeah, that's what I meant =) thx Feb 28 10:12:59 and damn my irc client Feb 28 10:13:06 it completes to tomimo instead of tom_ :) Feb 28 10:14:03 no problem with that ;) Feb 28 10:14:24 i have a amd x2 6000+ and a 4gb xen-vm running with 2 vcpus Feb 28 10:14:28 and make -j 4 Feb 28 10:14:46 but even then it takes quite some time Feb 28 10:15:22 it does Feb 28 10:15:30 but i have a second host with 2 cores and a 2gb vm, so i'll test icecc with that Feb 28 10:15:36 on core2 quad it flies nice Feb 28 10:15:41 hehe Feb 28 10:15:48 will have to buy something like that one day Feb 28 10:16:04 a quadcore intel with a softraid 0, BB_NUMBER_THREADS = "4" and PARALLEL_MAKE = "-j 8" builds everything from scratch (including toolchain, etc.) in about 40 minutes or so, maybe even less Feb 28 10:16:22 yeah now i know a *real* scenario for quadcore Feb 28 10:16:25 wooha Feb 28 10:16:31 :) Feb 28 10:16:41 i have hw raid1 Feb 28 10:16:50 whats BB_NUMBER_THREADS? Feb 28 10:17:19 * tom_ is just going for coffee Feb 28 10:17:58 tom_: I think its the number of things/tasks that bitbake will do at the same time Feb 28 10:18:00 tom_: amount of bitbake threads - set to 4 on dual core (or even more if want Feb 28 10:18:19 on my amd64 x2 4200+ I use 4 bbthreads and make -j4 Feb 28 10:18:33 on quad core I used 16 bbthreads and make -j8 Feb 28 10:18:55 too bad that I lack space on desk ;( I have amd64 3200+ cpu unused Feb 28 10:18:57 I should probably try to increase the bbthreads and compare Feb 28 10:19:27 ah ok thx Feb 28 10:19:33 hrw: that's no excuse, use it under desk or in some remote room? :> Feb 28 10:19:54 yeah IP exists ;) Feb 28 10:20:15 funny...bbthreads is not in the sample-conf Feb 28 10:21:57 Jin^eLD: no space under desk too Feb 28 10:22:03 and I have small flat Feb 28 10:22:06 * hrw -> afk Feb 28 10:33:42 k, icecc seems to be up & running Feb 28 10:40:12 good morning Feb 28 10:43:08 hey florian Feb 28 10:58:33 hmm..someone using icecc over internet (so, not lan, different subnets)? Feb 28 10:59:11 re **** ENDING LOGGING AT Thu Feb 28 10:59:56 2008